Skip to content

roadmap.md

Latest commit

 

History

History
77 lines (53 loc) · 5.47 KB

roadmap.md

File metadata and controls

77 lines (53 loc) · 5.47 KB

WordClaw Roadmap

WordClaw is evolving from a headless CMS into a secure governance runtime for AI agents. This roadmap tracks our progress, separates stable product features from incubating ideas, and provides clarity on where the ecosystem is heading.

🚀 Flagship Releases

We promote one major flagship feature per minor version release to ensure stability and focus.

v0.4: Reactive MCP Sessions (Current)

  • Goal: Let long-lived agents subscribe to runtime events instead of polling for content, workflow, and integration changes.
  • Status: Rolling out. Remote MCP sessions, reactive topic subscriptions, filter-aware recipes, task guidance follow-up recommendations, and discovery/status exposure are live on main. Remaining work is focused on hardening the reactive contract around the most valuable topics and demos, not inventing a second event system.
  • Documentation: See RFC 0025.

v0.3: Schema-Aware Media Assets (Shipped)

  • Goal: Make media files a first-class part of the core runtime instead of pushing agents toward ad hoc external URLs.
  • Status: Shipped. Local and S3-compatible storage, schema-level asset references, derivative variants, multipart and direct-provider upload flows, signed and entitlement-gated delivery, MCP/CLI tooling, supervisor asset controls, and restore/purge lifecycle are live on main.
  • Documentation: See RFC 0023.

v0.2: Production-Ready L402 Flows (Shipped)

  • Goal: Provide an end-to-end, secure payment path for machine-to-machine transactions.
  • Status: Shipped. Agents can now natively purchase and redeem Lightning network macaroons (L402) for premium content access.
  • Documentation: See L402 Protocol.

v0.1: Context-Aware Headless CMS (Shipped)

  • Goal: Establish the core structured content models, validation schemas, and API primitives.
  • Status: Shipped. Core entity framework, domains, API Key scopes, role management, schema-aware content references, field-aware content queries, grouped content projections, public write lanes, and TTL lifecycle archival for session-like content are stable.

📅 Near-Term Priorities

These features are currently in active development or polishing phases for upcoming minor releases.

1. Supervisor UI Design System and Consistency (RFC 0033)

  • Status: Partially implemented / hardening.
  • Description: The responsive shell, feedback primitives, shared tables, and broad route coverage are already live. Current work is reducing duplicate route-local UI code, standardizing page headers/cards/dialogs/filters/inspectors, and growing UI test coverage for shared operator flows.

2. Domain Administration in UI and Backend (RFC 0034)

  • Status: Proposed.
  • Description: Multi-domain isolation, domain creation, onboarding, and shell-level domain switching are already live. The next step is a dedicated domains workspace plus backend detail, summary, and update contracts so tenant administration, handoff, and lifecycle entry points stop being fragmented across the shell selector, API Keys page, and low-level routes.

3. Vector RAG Simplification

  • Status: Rolling out.
  • Description: Making native pgvector semantic search easier to enable. We now auto-detect OPENAI_API_KEY on startup, automatically generating embeddings for published content without requiring complex external pipeline deployments.

4. Additional Asset Providers (Optional follow-up)

  • Status: On demand.
  • Description: RFC 0023 is effectively shipped for the supported product path. Additional object-storage adapters beyond the current local and S3-compatible providers are now a demand-driven extension, not a core product gap.

5. Extensibility and Plugins (RFC 0022)

  • Status: Proposed.
  • Description: Designing a Strapi-style plugin architecture to allow the community to extend the Express/Fastify API boundaries and MCP toolsets without continuously forking the core runtime.

6. API Polish and Error Schemas

  • Status: In progress.
  • Description: Standardizing error payloads across both GraphQL and REST, ensuring AI agents receive consistent, actionable recommendedNextAction directives when validations fail.

7. Performance & Scalability (RFC 0024)

  • Status: Proposed.
  • Description: Establishing caching layers (Redis) for Agent capabilities/search, defining horizontal node scaling strategies, and targeting 100 concurrent agent runs at under 200ms latency.

🧪 Incubating / Experimental

These features are turned OFF by default in production. They represent longer-term architectural bets and agentic economy experiments. They may change drastically or be removed based on feedback.

AP2 Agentic Monetization (RFC 0016)

Testing advanced revenue splitting between model providers, system operators, and original content authors based on LLM consumption traces.

Internal Agentic Content Recommender

Exploring whether autonomous background agents can effectively curate and tag content relationships better than manual human curation.

Delegation and Worker Sandboxes

Testing secure boundaries for executing third-party or multi-step agent workflows via queued AgentRuns.

Note to Contributors: We recommend building tools and integrations against the Tier 1 Feature constraints defined in the Features Concept Guide rather than relying on experimental modules.